J. Sleep Res.

(2006) 15, 117–124

Evaluating the awakening criterion in the definition

of nightmares: how certain are people in judging whether
a nightmare woke them up?
M A R K B L A G R O V E and S U Z A N N E H A Y W O O D
Department of Psychology, University of Wales Swansea, Swansea, UK

Accepted in revised form 30 November 2005; received 29 November 2004

SUMMARY There is debate about whether to include in the definition of nightmares a criterion that
the imagery or emotions of the nightmare caused the person to wake up. This study
investigates whether people believe that they can judge this cause of awakening. 42
participants recorded for 14 nights whether they had a dream, and decided for each
dream whether it had or had not woken them. They then rated on a 5-point scale (where
1 ¼ very certain and 5 ¼ very uncertain) how certain they were in their decision of
whether or not the dream woke them. ParticipantsÕ mean certainty was high for
decisions that the dream woke them (mean certainty ¼ 1.60), and for very unpleasant
dreams this mean certainty that the dream woke them was very high (mean
certainty ¼ 1.27). Dreams judged to have caused awakening were found to be more
unpleasant than dreams judged not to have caused awakening. Although the inclusion
of the awakening criterion did not increase the association of nightmare frequency with
anxiety, there may be other advantages in the use of the awakening criterion.
keywords anxiety, awakening, definition, nightmares

content and an elaborated storyÕ (Wood et al., 1992), and

INTRODUCTION
There is debate about whether to include in the definition of
nightmares the criterion that the person was woken by the
events or emotions of the nightmare. For example, Levin and
Fireman (2002a), following the Diagnostic and Statistical
Manual of Mental Disorders, 4th edition (American Psychi-
atric Association, 1994) and the International Classification of
Sleep Disorders-R (American Sleep Disorders Association,
1997), define idiopathic nightmares as Ôvivid dreams marked by
intensified feelings of dread or terror that awaken the
individualÕ. Similarly, Zadra and Donderi (2000) define a
nightmare as Ôa very disturbing dream in which the unpleasant
visual imagery and/or emotions wake the person upÕ. In
contrast, other definitions of nightmares do not require the
person to have been woken by the nightmare: examples of
these definitions include Ôfrightening dreams with visual
Correspondence: Mark Blagrove, Department of Psychology, Univer-
sity of Wales Swansea, Swansea SA2 8PP, UK. Tel.: +44 1792 295586;
fax: +44 1792 295679; e-mail: m.t.blagrove@swansea.ac.uk
Ôvery disturbing dreams, involving any unpleasant emotion,
which are usually vividly recalledÕ (Belicki, 1992a). In this
paper the former category of nightmares will be termed waking
criterion nightmares and the latter category will be termed
inclusive definition nightmares. The category of inclusive
definition nightmares thus includes waking criterion night-
mares and bad dreams: in bad dreams, as a matter of definition
(see Zadra and Donderi, 2000), the events or emotions of the
dream are not the cause of the dreamer waking up.
Awakenings due to nightmares are common. Zadra and
Donderi (2000) found that 47% of their participants in a
1-month prospective log study reported having at least one
waking criterion nightmare. Their log data gave for their
sample an estimated yearly mean frequency of 11.04 for
waking criterion nightmares and 29.40 for bad dreams.
However, individuals can be inaccurate in their judgements
about sleep, such as about whether one was awake or asleep
when cued (Mendelson, 1998) and remembering brief experi-
mentally induced arousals (Kato et al., 2004). This problem of
subjective reports may be present also in expecting people to

Ó 2006 European Sleep Research Society 117

118 M. Blagrove and S. Haywood
judge that a nightmare woke them. For example, there may be
a halo effect when the person decides whether the nightmare
woke them, with awakening being erroneously attributed to a
dream that had especially unpleasant events and emotions.
Further problems with the waking criterion are that it may be
difficult to accurately introspect about what caused an
awakening, unless the cause is obvious, such as an alarm
clock. Participants claiming that the dream woke them may be
Ôtelling more than we can knowÕ (Nisbett and Wilson, 1977).
However, contrary to this sceptical approach to introspection
is the work of Ericsson and Simon (1993), who show that
immediate retrospective descriptions of specific heeded behav-
iours can be highly accurate.
Using such introspective reports Suengas and Johnson
(1988) found that memories for perceived events (for which
they include dreaming as one example) have greater clarity
and sensory and contextual detail than do memories for events
that the person had been told to imagine, and they conclude
that therefore people can engage in accurate reality monito-
ring, that is, determining whether a memory is of an event that
had been perceived. Johnson (1988) reviews how time (on the
scale of 24 or 48 h) and rehearsal can adversely affect reality
monitoring of perceptual and autobiographical memories, and
for us this has implications for the long-term retrospective
recall of nightmares, but this work does indicate that, upon
waking, people may be able to correctly judge whether or not
a dream woke them. In support of this is the finding of Zadra
and Donderi (2000) that out of their 42 participants who had
at least one waking criterion nightmare during their 1-month
study, only one reported difficulties in distinguishing awaken-
ing caused by the dream content from spontaneous or
externally caused awakenings.
Our primary aim in the present study was to extend this
finding of Zadra and Donderi (2000) by ascertaining how
certain people are in deciding that a nightmare woke them.
This method follows from Ericsson and Simon’s (1993) review
of the positive relationship between confidence and accuracy.
To reduce halo effects and attributions, decisions about
awakenings were made for all dreams, whether pleasant or
unpleasant, and certainty was assessed for decisions that a
dream had woken them, and also for decisions that a dream
had not woken them.
Our second aim was to (i) replicate the finding of Zadra
and Donderi (2003) that the mean hedonic tone of waking
criterion nightmares is more unpleasant than the mean
hedonic tone of bad dreams, and (ii) to extend this
comparison to dreams in general, by assessing whether
dreams that cause awakening are more unpleasant than
dreams that do not cause awakening.
Our third aim was separate from the issue of the validity of
the waking criterion, and concerned testing a possible
advantage of the criterion. This hypothesized advantage is
that the frequency of waking criterion nightmares will have a
higher correlation with waking life psychopathology or poor
well-being than would the frequency of inclusive definition
nightmares: data and theoretical reasons in favour of this
advantage are presented by Zadra and Donderi (2000). We
aimed to test this hypothesized advantage by correlating
frequencies of waking criterion and inclusive definition night-
mares with trait and state anxiety.
One of the few studies to have looked at correlates of
awakenings from dreams, Kroth et al. (1997), did find that a
combined anxiety–depression measure correlated significantly
(r ¼ 0.42) with a retrospective measure of Ôawakening abruptly
from a dreamÕ, but it is not clear whether the result was
confounded by the emotional tone of dreams. Furthermore, in
the literature in general, there is no systematic difference
between correlations of anxiety with frequencies of waking
criterion and inclusive definition nightmares. A survey of, to
our knowledge, all such between-subjects correlation studies in
the literature involving anxiety and nightmare frequency,
categorized by use or non-use of the awakening criterion, is
seen in Table 1. Similarly, in non-correlational studies, signi-
ficant relationships between nightmare frequency and trait
anxiety have been found by Hersen (1971), Levin (1998) and
Ohayon et al. (1997) using the waking criterion, and by
Mindell and Barrett (2002), Nielsen et al. (2000) and Tanska-
nen et al. (2001) without the waking criterion. However, a
concern with these correlational and non-correlational studies
of waking criterion nightmares is that participants did not
assess whether the nightmare woke them separately from
assessing the degree of unpleasantness or distress of the
nightmare: in the present study participants rate unpleasant-
ness of dreams and judge whether the dream woke them as
separate decisions.
METHOD
Participants
We recruited by advertisement 42 participants from the
university population (33 females, nine males, mean
age ¼ 21.79 years, SD ¼ 6.55) who were reporting at least
one nightmare per month. Local ethics committee approval
was obtained for the study and written consent to take part
was given by all participants.
Materials and procedure
Participants completed the State–Trait Anxiety Inventory
(Spielberger, 1983) to obtain state anxiety (Ôhow you feel
right now, that is, at this momentÕ) and trait anxiety (Ôhow
you generally feelÕ). They next completed the Nightmare
Distress Scale (Belicki, 1992b), a 13-item questionnaire on
which higher scores indicate a greater distress caused by
nightmares, such as in making one feel afraid of going to sleep
or wishing to have therapy to stop the nightmares. They next
made a retrospective estimate of their frequency of dream
recall and inclusive definition nightmares using the definition:
Ôvery disturbing dreams, involving any unpleasant emotion,
which are usually vividly recalledÕ. They then estimated their
frequency of waking criterion nightmares: the question here
Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124
 Judging nightmare awakening 119

Table 1 Pearson and Spearman correlation coefficients between nightmare frequency and trait and state anxiety

Category of
Retrospective nightmare or Trait State
Authors /prospective bad dream anxiety anxiety

Blagrove et al., 2004 R W 0.16

Zadra and Donderi, 2000 R W 0.27 and 0.37a 0.28 and 0.35a
(Starker and Hasenfeld, 1976) R W (0.25)b
Cook et al., 1990 R I 0.33
Haynes and Mooney, 1975 R I 0.17 and 0.23c
Krakow et al., 2002 R I 0.29 and 0.28d
Miró and Martı́nez, 2005 R I 0.23 and 0.06e
Nguyen et al., 2002 R I 0.26 and 0.19f
Schredl et al., 1996 R I 0.31
Wood and Bootzin, 1990 R I 0.13
Zadra and Donderi, 2000 R BD 0.16 and 0.05a 0.09 and 0.09a
Blagrove et al., 2004 P– 2 weeks W 0.12
Levin and Fireman, 2002a P– 3 weeks W 0.23 0.20
Zadra and Donderi, 2000 P– 1 month W 0.28 0.27
Zadra et al., 1995 P– 2 weeks W 0.23
Blagrove et al., 2004 P– 2 weeks I 0.35
Krakow et al., 2002 P– 2 weeks I 0.33 and 0.37d
Wood and Bootzin, 1990 P– 2 weeks I 0.04
Wood et al., 1992 P– 3 weeks I 0.29, 0.38 and 0.23g
Wood et al., 1992 P– 3 weeks I All 3 correlations
non-significanth
Zadra and Donderi, 2000 P – 1 month BD 0.09 0.08
Zadra et al., 1995 P – 2 weeks BD 0.31

Studies are categorized as retrospective (R) or prospective (P); nightmares categorized as using the waking criterion (W) or as inclusive definition
nightmares (I). Studies of frequency of bad dreams, where, by definition it was not the dream that caused the person to wake, are labelled BD.
Studies using inclusive definition nightmares thus combine waking criterion nightmares and bad dreams.
a
First coefficient is for nightmare frequency assessed for 1 year retrospectively and second coefficient is for nightmare frequency assessed for
1 month retrospectively.
b
Measure used for correlation was anxious-distractible daydreaming style. As this is not a standard trait anxiety measure citation and coefficients
are placed in parentheses.
c
Coefficients are from study I and study II.
d
First coefficient refers to nights per week that a nightmare occurred and second coefficient refers to number of nightmares per week.
e
First coefficient refers to frequency of nightmares for participants who have nightmares on a weekly basis, i.e. 1–7 times per week and second
coefficient refers to frequency of nightmares for participants who have nightmares on a monthly basis, i.e. 1–3 times per month.
f
First coefficient is for global anxiety and second coefficient is for anxiety related to testing.
g
Coefficients reported for three groups; first two correlations are for individuals who were in the proximity of the 1989 San Francisco earthquake
and third correlation is for a non-earthquake control group. Anxiety assessed for time of occurrence of earthquake.
h
Coefficients not reported due to being non-significant for each of the three groups tested. Anxiety assessed for the 24 h before the log started.

was ÔHow often do you have a nightmare where the emotion or
imagery of the nightmare wakes you up?Õ
Participants then kept a prospective dream log for 14 nights,
each morning answering the following:
1 Did you have a dream last night?
2 How pleasant/unpleasant was the dream? [responses made
on the 7-point dream hedonic tone scale (Foulkes et al.,
1966): 1 ¼ very pleasant to 7 ¼ very unpleasant]
3 Did the emotion of the dream wake you up?
4 Rate how certain you are of your answer to question 3
(responses made on a 5-point scale: 1 ¼ very certain to
5 ¼ very uncertain).
Participants could answer the questions for more than one
dream per night. Question 3 only referred to the emotion of the
dream causing waking, and did not refer to the possibility that
the imagery alone of the dream caused waking, unlike in the
fuller retrospective question about waking criterion night-
mares. It is not known whether or to what extent responses
would have altered if this question had also included reference
to dream imagery.
RESULTS
Table 2 shows the means, SDs and ranges of: retrospective
dream recall frequency; retrospective inclusive definition and
waking criterion nightmare frequencies; dream recall and
nightmare frequencies on the 2-week prospective log (log
inclusive definition nightmares defined as dreams rated 6 or 7,
i.e. moderately or very unpleasant, on the hedonic tone scale,
log waking criterion nightmares defined as dreams rated 6 or 7
on the hedonic tone scale where the person also states that they
were woken by the emotion of the dream); hedonic tone of the
log dreams; confidence in making decisions about whether or
not a dream caused an awakening; trait anxiety and state

Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124

120 M. Blagrove and S. Haywood

Table 2 Means, SDs and ranges of: retrospective dream recall fre- number of dreams (279.2), inclusive definition (53.3) and
quency (DRF); retrospective inclusive definition and waking criterion waking criterion nightmares (31.5) were higher than estimates
nightmare frequencies (NFs); dream recall and nightmare frequencies
from the retrospective questions (169.7, 35.9 and 20.1 respect-
on the 2 week prospective log (log inclusive definition nightmares de-
fined as dreams rated 6 or 7, i.e. moderately or very unpleasant, on the ively).
hedonic tone scale, log waking criterion nightmares defined as dreams There were no significant sex differences in anxiety, night-
rated 6 or 7 on the hedonic tone scale and where the person also states mare distress, retrospective or log dream recall frequency,
that they were woken by the emotion of the dream); hedonic tone of dream hedonic tone, certainty in decisions about dreams
the log dreams; confidence in making decisions about whether or not a
causing awakening or not causing awakening, or in frequency
dream caused an awakening; trait anxiety and state anxiety
of retrospective waking criterion nightmares, but males did
Variable Mean SD Range report significantly fewer retrospective inclusive definition, log
Retrospective DRF (per year) 169.67 105.17 30–286 inclusive definition and log awakening criterion nightmares
Retrospective inclusive definition NF 35.86a 49.16 10–286 than did females (inferential statistics in notes to Table 2).
(per year) Tests of normality showed only the anxiety variables to be
Retrospective waking criterion NF 20.05 29.39 0–104 normally distributed and so non-parametric inferential statis-
(per year)
tics are used for tests involving the other variables. The use of
Nightmare distress 25.50 7.66 13–55
Log DRF (per 2 weeks) 10.74 4.75 4–23 non-parametric statistics for the nightmare data was also
Log inclusive definition NF (per 2 weeks) 2.05b 1.91 0–9 appropriate given the inherent inaccuracies in retrospective
Log waking criterion NF (per 2 weeks) 1.21c 1.41 0–7 and brief log methodologies.
Hedonic tone of log dreams 3.89 0.89 2.00–6.33 Table 3 shows the results for all 42 participants of number
Confidence in all decisions of whether 1.94 0.58 1.00–3.35
and percentage of dreams in each hedonic tone category, and,
dream did or did not cause awakening
Trait anxiety 44.05 8.51 31–64 for each hedonic category, number and percentage of dreams
State anxiety 39.38 9.71 20–57 judged to have caused awakening, percentage judged not to
have caused awakening, mean certainty of judgements that
Note: Some dreams rated 6 might not have been termed a nightmare or
awakening was not caused by a dream and mean certainty of
bad dream by the participant, or might not have met some of the
definitions of nightmare or bad dream present in the literature. judgements that awakening was caused by a dream.
a
Males reported significantly fewer inclusive definition nightmares
[mean (SD) ¼ 20.44 (31.33) per year] on the retrospective question-
naire than did females [mean (SD) ¼ 40.06 (52.59) per year]; Mann– Aim 1
Whitney U ¼ 74.00, z ¼ 2.49, P ¼ 0.021.
b Observing Table 3 suggests that the judgements of being
Males reported significantly fewer log inclusive definition nightmares
[mean (SD) ¼ 0.89 (1.17)] over the 2 weeks than did females [mean woken by a dream are generally made with high certainty and
(SD) ¼ 2.36 (1.97)]; Mann–Whitney U ¼ 74.00, z ¼ 2.35, P ¼ 0.021. with very high certainty for the very unpleasant dreams.
c
Males reported significantly fewer log waking criterion nightmares Judgements that dreams did not cause awakening were made
[mean (SD) ¼ 0.44 (0.73)] over the 2 weeks than did females [mean with moderate to high certainty. To assess the statistical
(SD) ¼ 1.42 (1.48)]; Mann–Whitney U ¼ 79.50, z ¼ 2.24. P ¼ 0.033.
significance of the difference in confidence between decisions of
dreams causing awakening and of dreams not causing
anxiety. On the logs 35 participants (83.3% of the sample) awakening a subset of participants was used, which would
reported one or more inclusive definition nightmares and 28 allow within-subject comparisons and adequate sampling of
participants (66.7% of the sample) reported one or more those dreams judged to have caused, and those judged not to
waking criterion nightmares. As has been found previously have caused awakening. On the prospective dream log 18
(e.g. Wood and Bootzin, 1990; Zadra and Donderi, 2000), the participants had at least three instances of a dream that they
prospective log provided higher frequency estimates than did judged woke them and at least three instances of a dream that
the retrospective questions. Here, log estimates of yearly they judged did not wake them. These participants were

Table 3 Number and percentage of dreams categorized by hedonic tone, and, for each hedonic tone category, percentage of dreams judged to have
not caused awakening, number and percentage of dreams judged to have caused awakening, mean certainty of participants that a dream did not
cause awakening, and mean certainty of participants that a dream caused awakening (1 ¼ very certain, 5 ¼ very uncertain)

Hedonic tone (1 ¼ very pleasant, 7 ¼ very unpleasant)

1 2 3 4 5 6 7

Number of dreams with specified hedonic tone 36 65 81 119 64 52 34

Percentage of dreams with specified hedonic tone 8.0 14.4 18.0 26.4 14.2 11.5 7.5
Number of dreams judged to have caused awakening 9 5 12 8 16 24 26
Percentage of dreams judged not to have caused awakening 75.0 92.3 85.2 93.3 75.0 53.8 23.5
Percentage of dreams judged to have caused awakening 25.0 7.7 14.8 6.7 25.0 46.2 76.5
Mean certainty of judgements that a dream did not cause awakening 2.15 1.97 1.77 1.68 2.15 2.29 2.25
Mean certainty of judgements that a dream did cause awakening 1.44 1.80 2.08 1.62 2.31 2.08 1.27

Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124


Table 4 Kendall’s tau-b partial correlation coefficients between retrospective and log measures of inclusive definition and waking criterion
nightmare frequencies (NFs), trait nightmare distress and mean hedonic tone of all dreams, and correlations between all nightmare variables and
dream hedonic tone with trait and state anxiety
Retrospective Retrospective
inclusive waking
definition NF criterion NF
Retrospective waking criterion NF 0.57**
Log inclusive definition NF 0.31* 0.35**
Log waking criterion NF 0.34** 0.40**
Nightmare distress 0.21 0.31*
Mean dream hedonic tone 0.30* 0.12
Trait anxiety 0.21 0.11
State anxiety 0.24 0.10
Critical values of tau-b for P-values taken from Maghsoodloo and Pallos (1981). Sex is partialled out for all correlations
*P < 0.005; **P < 0.001.
significantly more certain in judgements that dreams had
woken them [mean certainty ¼ 1.60 (SD ¼ 0.62)] than in
judgements that dreams had not woken them [mean cer-
tainty ¼ 2.11 (SD ¼ 0.74), Wilcoxon z ¼ 2.37, P ¼ 0.018].
Mean certainty in judging that dreams caused the awakening
and mean certainty in judging that dreams did not cause
awakening each had negligible Spearman correlations with
state and trait anxiety (all |rsps | <0.08) and non-significant
correlations with nightmare distress (both |rsps | <0.15).
Aim 2a
Waking criterion nightmares were reported by 28 participants
with a mean hedonic tone of 6.47 (SD ¼ 0.44), and bad
dreams were reported by 21 participants with a mean hedonic
tone of 6.23 (SD ¼ 0.37). For inferential statistics we used the
participants who had at least one waking criterion nightmare
and at least one bad dream (n ¼ 14). For the waking criterion
nightmares, mean hedonic tone ¼ 6.48 (SD ¼ 0.43), which is
more unpleasant, but not significantly so, than for bad dreams,
where mean hedonic tone ¼ 6.23 (SD ¼ 0.37), Wilcoxon
z ¼ 1.65, P ¼ 0.10 (two-tail). Zadra and Donderi (2003) had
found a significant difference between waking criterion night-
mares where emotional intensity ¼ 7.95 (SD ¼ 1.51) and bad
dreams where emotional intensity ¼ 7.24 (SD ¼ 1.15), using a
9-point scale.
Aim 2b
Observing Table 3 suggests that more awakenings are judged
to be caused by very unpleasant dreams than by moderately
unpleasant dreams, and that moderately and very unpleasant
dreams are more likely to be judged to cause awakenings than
more pleasant dreams. We used the data from the 18
participants who had at least three instances of a dream that
they judged woke them and at least three instances of a dream
that they judged did not wake them, to investigate the
relationship between being woken from a dream and dream
pleasantness/unpleasantness. Here, dreams judged to have
caused awakening were found to be significantly more
Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124
Judging nightmare awakening 121
Log inclusive Log waking Nightmare Mean dream
definition NF criterion NF distress hedonic tone
0.64**
0.01 0.14
0.44** 0.43** )0.17
0.16 0.22 0.09 0.10
0.11 0.08 0.02 0.05
unpleasant [mean hedonic tone ¼ 4.89 (SD ¼ 1.33)], than
were dreams judged not to have caused awakening [mean
hedonic tone ¼ 3.70 (SD ¼ 0.88), Wilcoxon z ¼ 3.16,
P ¼ 0.002].
Aim 3
Finally, we used the data from all the participants for our third
aim, to investigate the relationships between the nightmare
variables and state and trait anxiety (the latter two being found
to correlate significantly, Pearson’s r ¼ 0.72, P < 0.001).
Table 4 shows the Kendall tau-b partial correlation coefficients
between the nightmare and hedonic tone variables, and
between them and trait and state anxiety, sex is partialled
out for all correlations. (Note should be taken that values of
Spearman’s r are often 50% higher than Kendall’s tau-b for
the same data and that Spearman and Kendall rank correla-
tion measures cannot be directly compared; Chen and Popov-
ich, 2002.) Following Bonferroni’s correction trait and state
anxiety did not correlate significantly with any nightmare
variable, nor with hedonic tone, and there were no systematic
differences between inclusive definition and waking criterion
nightmare frequencies in their correlation coefficients.
DISCUSSION
Regarding our first aim, individualsÕ judgements that the
emotions of a dream woke them were made with high certainty
and with very high certainty in the case of very unpleasant
dreams. Also, these judgements were made with significantly
more certainty than when judging that it was not the dream
that caused the awakening. If such certainty can be taken as
evidence that people are accurate when ascribing their
awakening to the dream, there may indeed be validity in
including the awakening criterion in the definition of night-
mares. This view is further supported by the greater unpleas-
antness/intensity of waking criterion nightmares in
comparison to bad dreams (Zadra and Donderi, 2003, and
our results for Aim 2a), and by our finding that dreams in
general that are judged to cause awakening are significantly
122 M. Blagrove and S. Haywood
more unpleasant than dreams judged not to cause awakening
(Aim 2b).
Importantly, level of certainty was not related to state or
trait anxiety or nightmare distress. Levin and Fireman (2002b)
had found nightmare prevalence to be more a function of trait
nightmare distress than of state nightmare intensity, calling
into question some decisions of the presence of a nightmare
because the decision is not simply a function of the actual
nightmare’s characteristics, but is rather due to heightened
sensitivity to its contents. Also, Edinger et al. (2000) found
that anxiety is predictive of erroneous judgements of sleep
quality. Our findings that certainty judgements about awa-
kening are not related to anxiety, or to trait nightmare distress,
do support the validity of these certainty judgements.
We also found little difference between correlations of
anxiety with frequencies of the two categories of nightmare
(Aim 3). This accords with the general conclusion from the
literature that the correlations between anxiety and nightmare
frequency, however nightmares are defined, are small and often
insignificant. However, caution is needed here because we did
not recruit control participants who report few or no night-
mares and this may have reduced the correlation coefficients of
either or both of the nightmare categories. Also, correlations
involving retrospective assessments may vary as a function of
dream recall frequency. Beaulieu-Prévost and Zadra (2005a)
found that the correlation of trait anxiety with retrospective
assessments of anxiety in dreams was significant for low
dream recall frequency participants (r ¼ 0.34), but negligible
(r ¼ 0.05) for high dream recall frequency participants. They
conclude that the higher correlation occurs due to the influence
of current affective state on beliefs about dream content for
individuals with limited access to instances of dreams on which
to base their beliefs about their dream content. This study also
found that the retrospective estimation of how much anxiety is
present in dreams has no significant relationship with pros-
pective measures in people with low dream recall frequency.
Our finding of there being apparently certain knowledge of
when nightmares cause awakenings, but with taking this
occurrence of awakenings into account not leading to an
augmentation of the correlation of nightmare frequency with
anxiety, indicates that such awakenings are not a function of
anxiety. Instead, being susceptible to being woken by a dream,
may, for example, be related to: ability to wake oneself up,
such waking being a defence-response, as suggested by Haynes
and Mooney (1975); measures of low well-being other than
anxiety, such as depression or daily hassles or recent negative
events (Cook et al., 1990); personality measures such as thin
intrapsychic boundaries (Hartmann, 1998) or absorption and
vividness of imagery (Belicki and Belicki, 1986); or individual
differences in the orienting reaction, either to dreamt or to
external (Fisher et al., 1970) stimuli. The personality and
orienting response explanations may indeed be more fitting
than psychopathology or low well-being based ones to address
our finding that 25% of very pleasant dreams caused awake-
nings, and that the mean certainty of these awakening
judgements was very high. Following the result of Zadra and
Donderi (2003), showing that waking criterion nightmares are
significantly more unpleasant than bad dreams, and our
similar but non-significant result for this comparison, it may
be this intensity level that causes awakening, but that intensity
level is not related to anxiety, but rather to some other low
well-being measure, or, again, to variables such as thin
boundaries or absorption, which are related to experiencing
nightmares (Belicki and Belicki, 1986; Hartmann, 1998).
However, although trait and state anxiety may only have
weak relationships with nightmare frequency in between-
subjects analyses, state anxiety is related to nightmare occur-
rence in the longitudinal study of Cellucci and Lawrence
(1978), where nightmares were defined as Ôa subjectively
disturbing or anxiety-provoking dream from which a person
usually awakensÕ, and in the waking criterion nightmare
treatment study of Kellner et al. (1992). There has not been,
though, any longitudinal study that addresses within-subject
correlations of both types of nightmare, or of bad dreams, with
state anxiety, or with any other well-being measure.
Following our investigation of this one possible advantage
of the waking criterion, we will now briefly mention four
others.
1 Does the use of the waking criterion aid nosology in
providing a more clear and exact definition of nightmares
and hence a more reliable measure? One issue here is whether
the two categories of nightmare differ in the accuracy of
retrospective frequency estimations. Such differences could
occur because of differences in the ease of making instances
of the two categories available to memory (Tversky and
Kahneman, 1974), which could be a function of the interest
in and attitude towards dreams and nightmares that the
person has (Beaulieu-Prévost and Zadra, 2005b).
2 Do waking criterion nightmares produce more distress on
average than do inclusive definition nightmares? As already
discussed, waking criterion nightmares are more unpleasant
than bad dreams, and so it may well be that the mean
emotional response following waking criterion nightmares
may be worse than for inclusive definition nightmares and for
bad dreams. Yet, although plausible, this has not been directly
addressed in the literature. State distress resulting from
individual nightmares has been assessed for waking criterion
nightmares (Levin and Fireman, 2002b; Ohayon et al., 1997)
and inclusive definition nightmares (Köthe and Pietrowski,
2001; Krakow et al., 2002), but not for both definitions of
nightmare in one study. However, the related measure of trait
nightmare distress has been found to have a higher correlation
with prospective frequency of waking criterion nightmares
(tau-b ¼ 0.14 in the present study, r ¼ 0.36 in Blagrove et al.,
2004, r ¼ 0.47 in Levin and Fireman, 2002b) than with
prospective frequency of inclusive definition nightmares
[tau-b ¼ 0.01 in the present study, r ¼ 0.25 in Blagrove et al.,
2004, r ¼ 0.17 (nights per week) and r ¼ 0.30 (nightmares per
week) in Krakow et al., 2002].
3 Are waking criterion nightmares and inclusive definition
nightmares differentially responsive to treatment? For
example, frequency of waking criterion nightmares is
Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124

decreased by relaxation and desensitization (Miller and
DiPilato, 1983) and by desensitization and rehearsal treat-
ments (Kellner et al., 1992), but the effect of these treat-
ments on nightmares defined without the criterion, or on
bad dreams, has not been reported.
4 Defining nightmares with the waking criterion and hence
making the distinction with bad dreams may be productive
for future research, as consistency in definitions is essential
for results to be comparable across studies, and as in making
this distinction it is then possible to explore the possible
differential correlates of both types of dream. Data from
both dream experiences can also be combined to be
compared with studies reporting only on inclusive definition
nightmares. However, data from studies that only use the
inclusive definition cannot be so usefully and fully compared
to data from studies that use the waking criterion and bad
dream categories.
Although there may be such advantages as those mentioned
above in the use of the waking criterion, a possible confound is
that frequent awakenings from sleep in general have been
reported by individuals with nightmares compared to non-
nightmare controls in studies of inpatients in a large state
hospital (Hersen, 1971), a heterogeneous group including
participants who had suffered trauma (Krakow et al., 1995)
and insomniacs (Ohayon et al., 1997). These three studies used
subjective reporting, but poorer sleep in general has also been
found in polysomnographical studies of people with trauma-
related nightmares (Germain and Nielsen, 2003; Woodward
et al., 2000). It may be that individuals in such clinical
populations are able to distinguish awakenings not caused by
nightmares from instances of awakenings caused by night-
mares, but as their frequent nocturnal awakenings would
sometimes occur at the same time as a nightmare, they may on
some occasions erroneously believe that the nightmare caused
the awakening. However, whereas there are findings for non-
clinical populations of subjective reports of increased night-
time awakenings for individuals with nightmares (Cellucci and
Lawrence, 1978; Levin, 1994; Schredl, 2003), this is not found
polysomnographically for idiopathic nightmares (Germain
and Nielsen, 2003).
In conclusion, the clinical use of the waking criterion is
supported by our finding that participants have high certainty
in judgements that a nightmare caused them to wake, as this
suggests they may thus be accurate when they judge that they
were awoken by the nightmare. However, the waking criterion
may be problematic for populations with trauma-related
nightmares, or those with primary insomnia, because of the
tendency of people in such groups to wake frequently from
sleep in general. Regarding one of the hypothesized advantages
of the waking criterion, the use of the criterion did not lead to
greater predictive power of nightmare frequency for trait or
state anxiety. However, caution over the latter correlational
results is required, as our sample size was small for such
analyses, we did not recruit control participants with few or no
nightmares, and the correlation coefficients were low, although
they were in accord with much of the previous literature.
Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124
Judging nightmare awakening 123
Although this hypothesized advantage of the waking criterion
as an index of low well-being is currently under debate, there
are other possible advantages that we have detailed that also
warrant investigation.
REFERENCES
American Psychiatric Association. Diagnosis and Statistical Manual of
Mental Disorders, 4th edn. APA, Washington, DC, 1994.
American Sleep Disorders Association. International Classification of
Sleep Disorders: Diagnosis and Coding Manual, Revised. ASDA,
Rochester, MN, 1997.
Beaulieu-Prévost, D. and Zadra, A. How dream recall frequency
shapes people’s beliefs about the content of their dreams. North Am.
J. Psychol., 2005a, 7: 253–264.
Beaulieu-Prévost, D. and Zadra, A. Dream recall frequency and
attitude towards dreams: a reinterpretation of the relation. Pers.
Individ. Dif., 2005b, 38: 919–927.
Belicki, K. Nightmare frequency versus nightmare distress: relations to
psychopathology and cognitive style. J. Abnorm. Psychol., 1992a,
101: 592–597.
Belicki, K. The relationship of nightmare frequency to nightmare
suffering with implications for treatment and research. Dreaming,
1992b, 2: 143–148.
Belicki, K. and Belicki, D. Predisposition for nightmares: a study of
hypnotic ability, vividness of imagery, and absorption. J. Clin.
Psychol., 1986, 42: 714–718.
Blagrove, M., Farmer, L. and Williams, E. The relationship of
nightmare frequency and nightmare distress to well-being. J. Sleep
Res., 2004, 13: 129–136.
Cellucci, A. J. and Lawrence, P. S. Individual differences in self-
reported sleep variable correlations among nightmare sufferers.
J. Clin. Psychol., 1978, 34: 721–725.
Chen, P. Y. and Popovich, P. M. Correlation: Parametric and
Nonparametric Measures. Sage, Thousand Oaks, CA, 2002.
Cook, C. A. L., Caplan, R. D. and Wolowitz, H. Nonwaking
responses to waking stressors: dreams and nightmares. J. Appl. Soc.
Psychol., 1990, 20: 199–226.
Edinger, J. D., Fins, A. I., Glenn, D. M., Sullivan, R. J. Jr, Bastian, L.
A., Marsh, G. R., Dailey, D., Hope, T. V., Young, M., Shaw, E. and
Vasilas, D. Insomnia and the eye of the beholder: are there clinical
markers of objective sleep disturbances among adults with and
without insomnia complaints? J. Consult. Clin. Psychol., 2000, 68:
586–593.
Ericsson, K. A. and Simon, H. A. Protocol Analysis: Verbal Reports as
Data (revised edition). MIT Press, Cambridge, MA, 1993.
Fisher, C., Byrne, J., Edwards, A. and Kahn, E. A psychophysiological
study of nightmares. J. Am. Psychoanal. Assoc., 1970, 18: 747–782.
Foulkes, D., Spear, P. S. and Symonds, J. D. Individual differences in
mental activity at sleep onset. J. Abnorm. Psychol., 1966, 71: 280–286.
Germain, A. and Nielsen, T. A. Sleep pathophysiology in posttrau-
matic stress disorder and idiopathic nightmare sufferers. Biol.
Psychiatry, 2003, 54: 1092–1098.
Hartmann, E. Personality and dreaming: boundary structure and
dream content. Dreaming, 1998, 8: 31–39.
Haynes, S. N. and Mooney, D. K. Nightmares: etiological, theoretical
and behavioral treatment considerations. Psychol. Rec., 1975, 25:
225–236.
Hersen, M. Personality characteristics of nightmare sufferers. J. Nerv.
Ment. Dis., 1971, 153: 27–31.
Johnson, M. K. Reality monitoring: an experimental phenomenolog-
ical approach. J. Exp. Psychol. Gen., 1988, 117: 390–394.
Kato, T., Montplaisir, J. Y. and Lavigne, G. J. Experimentally induced
arousals during sleep: a cross-modality matching paradigm. J. Sleep
Res., 2004, 13: 229–238.
124 M. Blagrove and S. Haywood
Kellner, R., Neidhardt, J., Krakow, B. and Pathak, D. Changes in
chronic nightmares after one session of desensitization or rehearsal
instructions. Am. J. Psychiatry, 1992, 149: 659–663.
Köthe, M. and Pietrowski, R. Behavioral effects of nightmares and
their correlations to personality patterns. Dreaming, 2001, 11: 43–52.
Krakow, B., Tandberg, D., Scriggins, L. and Barey, M. A controlled
comparison of self-rated sleep complaints in acute and chronic
nightmare sufferers. J. Nerv. Ment. Dis., 1995, 183: 623–627.
Krakow, B., Schrader, R., Tandberg, D., Hollifield, M., Koss, M. P.,
Yau, C. L. and Cheng, D. T. Nightmare frequency in sexual assault
survivors with PTSD. J. Anxiety Disord., 2002, 16: 175–190.
Kroth, J., Jensen, L. and Haraldsson, M. Correlations of splitting and
phobic anxiety with dreaming. Percept. Mot. Skills, 1997, 85: 333–
334.
Levin, R. Sleep and dream characteristics of frequent nightmare
subjects in a university population. Dreaming, 1994, 4: 127–137.
Levin, R. Nightmares and schizotypy. Psychiatry, 1998, 61: 206–216.
Levin, R. and Fireman, G. Nightmare prevalence, nightmare distress,
and self-reported psychological disturbance. Sleep, 2002a, 25: 205–
212.
Levin, R. and Fireman, G. Phenomenal qualities of nightmare
experience in a prospective study of college students. Dreaming,
2002b, 12: 109–120.
Maghsoodloo, S. and Pallos, L. L. Asymptotic behavior of Kendall’s
partial rank correlation coefficient and additional quartile estimates.
J. Stat. Comput. Simul., 1981, 13: 41–48.
Mendelson, W. B. Effects of time of night and sleep stage on
perception of sleep in subjects with sleep state misperception.
Psychobiology, 1998, 26: 73–78.
Miller, W. R. and DiPilato, M. Treatment of nightmares via relaxation
and desensitization: a controlled evaluation. J. Consult. Clin.
Psychol., 1983, 51: 870–877.
Mindell, J. A. and Barrett, K. M. Nightmares and anxiety in
elementary-aged children: is there a relationship? Child Care Health
Dev., 2002, 28: 317–322.
Miró, E. and Martı́nez, M. P. Affective and personality characteristics
in function of nightmare prevalence, nightmare distress, and
interference due to nightmares. Dreaming, 2005, 15: 89–105.
Nguyen, T. T., Madrid, S., Marquez, H. and Hicks, R. A. Nightmare
frequency, nightmare distress, and anxiety. Percept. Mot. Skills,
2002, 95: 219–225.
Nielsen, T., Laberge, L., Paquet, J., Tremblay, R. E., Vitaro, F. and
Montplaisir, J. Development of disturbing dreams during
adolescence and their relation to anxiety symptoms. Sleep, 2000,
23: 727–736.
Nisbett, R. E. and Wilson, T. D. Telling more than we can know:
verbal reports on mental processes. Psychol. Rev., 1977, 84: 231–259.
Ohayon, M. M., Morselli, P. L. and Guilleminault, C. Prevalence of
nightmares and their relationships to psychopathology and daytime
functioning in insomnia subjects. Sleep, 1997, 20: 340–348.
Schredl, M. Effects of trait and state factors on nightmare frequency.
Eur. Arch. Psychiatry Clin. Neurosci., 2003, 253: 241–247.
Schredl, M., Pallmer, R. and Montasser, A. Anxiety dreams in school-
aged children. Dreaming, 1996, 6: 265–270.
Spielberger, C.D. Manual for the State-Trait Anxiety Inventory.
Consulting Psychologists Press, Palo Alto, 1983.
Starker, S. and Hasenfeld, R. Daydream styles and sleep disturbance.
J. Nerv. Ment. Dis., 1976, 163: 391–400.
Suengas, A. G. and Johnson, M. K. Qualitative effects of rehearsal on
memories for perceived and imagined complex events. J. Exp.
Psychol. Gen., 1988, 117: 377–389.
Tanskanen, A., Tuomilehto, J., Viinamäki, H., Vartiainen, E.,
Lehtonen, J. and Puska, P. Nightmares as predictors of suicide.
Sleep, 2001, 24: 844–847.
Tversky, A. and Kahneman, D. Judgment under uncertainty: heuris-
tics and biases. Science, 1974, 185: 1124–1131.
Wood, J. M. and Bootzin, R. R. The prevalence of nightmares and
their independence from anxiety. J. Abnorm. Psychol., 1990, 99: 64–
68.
Wood, J. M., Bootzin, R. R., Rosenhan, D., Nolen-Hoeksema, S. and
Jourden, F. Effects of the 1989 San Francisco earthquake on
frequency and content of nightmares. J. Abnorm. Psychol., 1992,
101: 219–234.
Woodward, S. H., Arsenault, N. J., Murray, C. and Bliwise, D. L.
Laboratory sleep correlates of nightmare complaint in PTSD
inpatients. Biol. Psychiatry, 2000, 48: 1081–1087.
Zadra, A. and Donderi, D. C. Nightmares and bad dreams: their
prevalence and relationship to well-being. J. Abnorm. Psychol., 2000,
109: 273–281.
Zadra, A. and Donderi, D. C. Affective content and intensity of
nightmares and bad dreams. Sleep (abstr suppl), 2003, 26: A93–A94.
Zadra, A. L., Assaad, J. -M., Nielson, T. A. and Donderi, D. C. Trait
anxiety and its relation to nightmares, bad dreams and dream
content. Sleep Res., 1995, 24: 150.
Ó 2006 European Sleep Research Society, J. Sleep Res., 15, 117–124